Table 2. Weighted MSE of all models, based on 5-features and 18-features inputs.

	5 features		18 features
	Train	Test	Train	Test
Constant prediction	1.61	1.62	1.61	1.62
Linear Regression	1.20	1.20	1.13	1.14
Linear regression + L1	1.17	1.17	1.12	1.12
Linear regression + L2	1.18	1.19	1.11	1.12
Linear regression + L1 + L2	1.17	1.16	1.11	1.11
Grad boosting 100 estimators	1.11	1.13	1.08	1.10
Grad boosting 250 estimators	1.06	1.11	0.95	1.07
biLSTM 64 units & 6 bins	0.83	0.88	0.79	0.84